Multi-tenant Data Processing Resource Rotation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In multi-tenant systems, efficiently distributing resources among multiple data models executed by a single application instance is challenging due to varying utilization patterns and demands, leading to suboptimal resource allocation and potential bottlenecks.

Innovation Solution

Implementing a system that automatically distributes resources based on a set of rotation factors, which are recalculated periodically using historical and current data, to ensure optimal allocation and real-time execution of data models, utilizing techniques such as weighted and lagged rotation data to adjust resource distribution dynamically.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a single application instance executes multiple data models simultaneously, then resource utilization efficiency is improved, but resource allocation optimization deteriorates due to varying utilization patterns

Engineering Contradiction:

Improveresource utilization efficiencyVSAvoidresource allocation optimization

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic resource allocation by continuously monitoring utilization patterns of multiple data models and adjusting resource distribution in real-time. The system transitions from static resource allocation to dynamic allocation based on current demand, allowing the single application instance to adapt to varying utilization patterns of different data models while maintaining high resource utilization efficiency.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes allocation parameters based on observed utilization patterns. By monitoring metrics such as execution frequency, resource consumption, and performance characteristics of each data model, the system adjusts resource allocation parameters dynamically, enabling optimal resource distribution across multiple data models executed by a single application instance.

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If resources are statically allocated among data models, then system complexity is reduced, but system performance deteriorates due to varying demands

Engineering Contradiction:

Improveresource allocation mechanismVSAvoidsystem performance

Core Design Contradiction:

Device complexityVSProductivity

Solution Approach 1:

The system implements self-service resource allocation where the resource management mechanism automatically monitors utilization patterns and adjusts resource distribution without external intervention. This autonomous adaptation allows the system to maintain simple architecture while achieving high performance through automatic resource optimization based on real-time demand characteristics of different data models.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent incorporates feedback mechanisms that continuously monitor the performance and utilization of each data model. This feedback information is used to dynamically adjust resource allocation, creating a closed-loop control system that automatically optimizes resource distribution based on actual system conditions, thereby maintaining high performance without increasing structural complexity.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If rotation factors are recalculated frequently, then resource allocation accuracy is improved, but computational overhead increases

Engineering Contradiction:

Improveresource allocation accuracyVSAvoidcomputational overhead

Core Design Contradiction:

Measurement precisionVSUse of energy by stationary object

Solution Approach 1:

The system implements periodic recalculation of rotation factors at strategically determined intervals rather than continuously. By balancing the frequency of recalculation with the rate of change in utilization patterns, the system achieves accurate resource allocation while minimizing unnecessary computational overhead. The periodic action is optimized to recalculate only when significant changes in demand patterns are detected.

Inventive Principle:

Principle #19Periodic action

Data Source

PatentUS9195513B2Systems and methods for multi-tenancy data processing

Publication Date: 2015.11.24 SAS INSTITUTE INC

AI summary

System and methods are provided for rotating real time execution of data models using an application instance. Input data are received for real time execution of a plurality of data models. An application instance is assigned for executing the plurality of data models simultaneously. Resources of the application instance are automatically distributed based on a set of rotation factors. The plurality of data models are executed simultaneously using one or more data processors. Execution results for one or more of the plurality of data models are output.